Although BOLD-fMRI is widely used for in-vivo functional brain mapping, there has been continuing interest in the possibility of using MRI to directly detect the neuronal activity. However, previous works failed to reach a clear consensus on that possibility. Here, we proposed a novel method for direct detection and spatial mapping of neuronal activity with temporal resolution on the order of the action-potential duration(4~5ms). We demonstrated its feasibility using an ex-vivo squid axon and an insulated copper-wire in an event-related electrical-stimulation paradigm. The results are very exciting and promising, although the underlying contrast mechanism needs to be elucidated.
Line-scan-based UTE Imaging Method: The main idea of the new method is that, if we could increase the temporal resolution up to the order of action-potential duration in an event-related design, transient effects of action potential could maximally be caught up and coherently accumulated in repeated trials, thereby being highly probable to be detected. For that purpose, we combined line-scanning method12 with 2D-UTE imaging, the schematic diagram of which is shown in Fig.1. At each trial, same k-space lines were acquired and trials were repeated incrementing the radial-view order. Temporal resolution was 4.5ms.
Data Processing and Analysis: Images were reconstructed with a home-built MATLAB(ver.8.2.0;R2013b) program using fast-Fourier-transform(FFT) with gridding. A series of 2D spatial mapping of responses to action potential or currents were acquired from the time-series images after filtering the time-course data. A notch filter was applied to eliminate 60-Hz electrical noise and a weak low-pass filter with a cutoff at 80Hz was applied to reduce noises.
Experiments: Experiments were performed on a 9.4-T Bruker (BioSpec 94/30, Ettlingen, Germany) scanner. A squid axon and a thin-copper wire were inserted into a cylindrical phantom with buffer solution and 1-% agarose gel, respectively, and placed perpendicular to B0 inside the magnet. A single coronal-slice was taken. Common scan parameters were: TR/TE = 4.5/0.25ms and, thus, temporal resolution = 4.5ms, slice-thickness = 5mm, Number-of-radial-spokes = 202. No averaging was tried.
Ex-vivo squid giant-axon imaging: An event-related design was used with three consecutive periods at each trial: 900ms(off)-0.1ms(on)-544.4ms(off). Electrical stimulation was applied with 110 mV, which is on the order of action-potential voltage. A series of 321 images were acquired: number-of-trials = 202, scan-time = 10min, spatial-resolution = 1.26mm2, FA=3°.
Current-carrying wire imaging: Each trial consisted of 1,800ms(off)-900ms(on)-900ms(off). Current was applied with 500mV. A series of 800 images were acquired: scan-time = 12min, spatial-resolution = 1.56mm2. FA = 3° and 20°.
[1] Bodurka J, Bandettini PA. Magn Reson Med. 2002; 47:1052-1058.
[2] Bianciardi M, et al. Magn Reson Imaging. 2004; 22:1429-1440.
[3] Kamei H, et al. IEEE Trans On Magnetics. 1999; 35:4109-4111.
[4] Chow LS, et al. NeuroImage. 2006; 30:835-846.
[5] Park TS, et al. Physiol Meas. 2006; 27:191-190.
[6] Petridou N, et al. Proc Natl Acad Sci. 2006; 103(43):16015-16020.
[7] Singh M. IEEE Trans Nucl Sci. 1994; 41:349-351.
[8] Chu R, et al. NeuroImage. 2004; 23:1059-1067.
[9] Parkes LM, et al. Magn Reson Med. 2007; 57:411-416.
[10] Tang L, et al. Magn Reson Imaging. 2008; 26:484-489.
[11] Luo Q, et al. NeuroImage. 2009; 47:1268-1276.
[12] Yu X, et al. Nat Methods. 2014; 11:55-58.